Tobias Scheer University of Nice, CNRS 6039
[email protected]
1
OCP 4, Rhodes 18-21 January 2007 this poster and many of the references quoted at www.unice.fr/dsl/tobias.htm
Mark€ta Zikov• University of Brno
[email protected]
(Slavic) vowel-zero alternations and phase theory 2
Havl€k vs. Lower
(1) all vowel-zero alternations share
(3) Havl€k's Law: illustration
a.
zero in open syllables (Cz lok€t-e)
"with the dog"
b.
vowel in closed syllables (Cz loket, loket-n€)
4 3 2 1 4 2 CS sъ pьs-ъmь > ocz se p•s-em• 5 4 3 2 1 4 2 CS sъ šьv-ьc-ьmь > ocz s• šev-•c-em•
"with the shoemaker"
Scheer (2004)
Havl€k
regular vowel-zero alternations in CVCV
Lower
Moroccan Arabic Russian German Modern Polish French Modern Czech Old Polish Old Czech
(5) analysis of vowel-zero alternations in CVCV (Scheer 2004,2005)
(4) Havl€k vs. Lower: illustration (2) they fall into two patterns according to how sequences of alternating vowels behave: a.
Cz Po
Lower vocalises all of them.
C__C-yer C•
C__C-•
C__C-CV
dom-•k-u
ocz dom- • č-ek mcz dom- e č-ek
dom-ek
dom-eč-•k-u
opol mpol
pies
p•s-a
p • s-ek p ie s-ek
lexical ingredients
Gvt
Havl€k: alternating vowels = floating melodies can govern
Gvt
Lower: alternating vowels = floating melodies cannot govern
C V C V C V C V |
|
|
|
observation: stable (i.e. non-alternating) suffix-initial vowels produce unvocalized alternation sites)
C V C V C V -
but they must lexically float for independent reasons: they must enter the final empty Nucleus.
C V C V C V
C V C V C V C V
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
l
o k e
t
e
l
o k e
t
l
o k e
t
e n
€
a. lokt-e GENsg
|
|
|
|
5
Gvt
C V
|
|
|
s
k e
l
o
Gvt
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
l
o k e
t
e
l
o k e
t
l
o k e
t
e n
€
solution 1 – alternating vs. non-alternating vowels
C V C V C V |
|
a.
s
k
the difference is the sensitivity to Government: alternating vowels care for Government, stable vowels do not
b.
stable vowels: if not floating, are lexically associated if floating, associate to any nucleus available no matter whether it is governed
c.
alternating vowels: always float lexically can only associate to ungoverned Nuclei
| e
l
•
root
GENpl
Gvt
Gvt
C V C V C V
C V C V C V
C V
C V
|
|
|
|
|
|
s
k e
l
t
†
-
o v
-
i
|
|
s
k
| e
l
-
root
o
alternating vowels only associate to ungoverned Nuclei
NOMsg
d.
and they needed “reduction”, i.e. to cut out empty VC units from the syllabic string. Reduction occurred only with vowel-zero alternations of the Lower type, nowhere else in the grammar.
C V
|
|
|
e k
-
C V C V C V
C V
C V
|
|
|
|
|
|
|
|
|
e k
s
k e
l
t
†
|
|
e k
|
C V C V
C V
|
|
|
|
d o m
-
Gvt
Gvt
e k
b. after reduction, input into cycle two
O N O N | | [
|
|
O N O N |
[ x x x x ]
|
|
x x x ]
| |
| |
|
|
| |
|
|
p
s
k
p ie s
k
•
10
C V
C V
|
|
|
|
|
d o m conclusion
-
e k
-
e k
phase boundary
|
[ x x x x x x ]
|
epenthesis
C V C V
O N O N O N
|
ie
9
PG
(12) conclusion a.
e epenthesis
b.
i
solution 2 – Havl€k vs. Lower
e. fair question: why is it that only suffixes with alternating vowels can be cyclic? Why should a phonological property determine the cyclic behaviour of its host? ==> phases are phonology-driven, cf. below.
Gussmann & Kaye (1993): derivation of Polish pies-ek
PG
-
d. suffixes with a stable vowel are always non-cyclic, i.e. sit in the same phase as the preceding morpheme.
Phase impenetrability: the association created in phase 2 cannot be undone
Gvt
Reduction
o v
c. yer-initial suffixes may or may not be cyclic Havl€k: they are not = sit in the same phase as the preceding morpheme Lower: they are = do not sit in the same phase as the preceding morpheme
phase 3
phase 2
|
d o m
-
b. hence: a given affix has exactly the same representation in Havl€k and Lower systems.
Mod Cz dom-eč-ek “house double dim”
|
-
a. Havl€k vs. Lower has a procedural, rather than a representational encoding.
e k
C V C V
e k
(9) solution 2: phase-sensitive affixes Havl€k vs. Lower
as many phases as there are suffixes
phase 1
-
6
Lower: cyclic suffixes
reduction is not needed in out analysis because unlike Gussmann & Kaye, we make a difference between floating and stable vowels: all suffix-initial vowels float (in Czech).
Gvt
C V
C V
C V -
Gvt
suffix –ek “diminutive”
Gvt
C V C V d o m
Gvt C V
| |
Cz sk•l-ov-it† “glassily”
C V C V d o m
OCz dom-•č-ek “house double dim” Gvt
stable vowels associate even to governed Nuclei
OCz dom-•č-ek “house double dim”
Havl€k: non-cyclic suffixes 1 single phase covers the stem and all affixes
they did not oppose Lower to Havl€k, though.
Gvt
C V C V C V C V
Cz sk•l-ov-it† “glassily”
|
Gvt
(8) solution 1: only alternating vowels care for Gvt alternating vs. non-alternating vowels
-o is NOT a yer: it remains stable
Gussmann & Kaye (1993) the solution favoured here is Gussmann & Kaye’s (1993). In their vocabulary, -ek is analytic (=cyclic) in Lower systems, and alternating vowels in a row survive because of the robustness of analytic domains.
Gvt C V C V C V
Cz sk•l-o “glass(NOMsg)”
diachronically, thus, all that has happened between OCz and ModCz is that -ek has become cyclic.
c. loket-n€ adjective
C V C V C V
NOMsg
where we stand now
b. loket NOMsg
Gvt
Gvt
Cz skel “glass (GENpl)”
|
root
hence suffix-initial vowels must float, but some of them are also stable, i.e. non-yers. This then is in contradiction with the definition of alternating vowels as floating melodies.
a. after the first cycle, before reduction
Government is an association-inhibitor
C V C V C V
C V C V C V C V
Gvt
c.
e.
e.
|
Cz *skel-o “glass(NOMsg)”
(10) summary
b.
they attach whenever they can
Czech "elbow"
Gvt
a.
d.
pies-•k-a
Gvt
suffixal-initial vowels must float
(7) problem: suffix-initial vowels must float, but do not alternate with zero
7
alternating vowels are floating pieces of melody
c. loket-n€ adjective
Mod Cz dom-eč-ek
problem with affixes
c.
c.
b. loket NOMsg
d o m e č e k
b.
empty Nuclei do not have any melody
a. lokt-e GENsg
Lower
|
a.
b.
Old Cz dom-•č-ek
d o m e č e k
4
full vowels are pieces of melody which are underlyingly attached
Havl€k
(6) implementation of Havl€k vs. Lower
b.
vowel
C__C-V
Havl€k vs. Lower in CVCV
a.
closed syllable
zero
Havl€k vocalises every other, counting from the right margin.
b.
3
open syllable
a.
a prediction is made: since the existence of a phase supposes concatenation, vowel-zero alternations within morphemes must always follow Havl€k. It is not so easy to find languages with more than one alternation site within a single morpheme, but all cases that we are aware of work: French (e.g. devenir), Moroccan Arabic (e.g. k•tɨb). fair question: why is it that only suffixes with alternating vowels are cyclic (in Slavic)? Why should a phonological property determine the cyclic behaviour of its host?
c.
floating vowels rather than Reduction.
d.
interpretation-driven phase theory: - are there really any interpretation-independent phases? - node-driven phase theory is blind to interpretation and faces serious trouble.
8
(11) Phase theory a.
phase-triggering as a property of affixes (or affix-classes) is an idea first put forth by Halle & Vergnaud (1987).
b.
hence phases are phonology-driven: there is one when we observe its phonological effects.
c.
ideally, the phonological traces of a phase coincide with morphological and/or syntactic properties of affixes. This is the original Lexical Phonology generalisation called affix ordering.
d.
a completely different perspective is “node-driven phase”: Marvin (2002) says that phases are triggered at every xP. Piggott (2006), Piggott & Newell (ms) also follow this track: he distinguishes strong (DP, CP) and weak (xP) phases.
e.
node-driven phase is certainly a desirable thing to have, but it appears to fall foul of the most basic and best known generalisations regarding English class 1 – class 2 morphology: origin-‚l-ity has two suffix xPs, but not a single phase is triggered – otherwise stress would not be penultimate (cf. par‰n-tal vs. pŠrent-hood).
f.
pejs-ek vs. ps-€k “dog dim (both)” if alternating vowels sit in a phase of their own in Lower systems, this example shows that there are phases which owe nothing to morpho-syntax at all: in absence of contrasting behaviour, -ek and -€k realize the same pieces of the morphosyntactic structure. The only difference is phonological: the initial vowel of -ek, but not of -€k, alternates with zero.
g.
phase theory is still at an embryotic level of development. One thing that needs to be sorted out, for instance, is the question whether there is any need for phases without traces in the interpretative modules at all.
h.
our best guess: phases exist for reasons of interpretation (Chomsky: UG reduces to merge and phase), hence
references
Gussmann, Edmund & Jonathan Kaye 1993. Polish notes from a Dubrovnik Caf€: I. The yers. SOAS Working Papers in Linguistics and Phonetics 3, 427-462. Halle, Morris & Jean-Roger Vergnaud 1987. An Essay on Stress. Cambridge, Mass.: MIT Press. Marvin, Tatjana 2002. Topics in the Stress and Syntax of Words. Ph.D dissertation, MIT. Piggott, Glyne 2006. The phonological inerpretation of morphological structure. Course handout, EGG summer school, Olomouc. Piggott, Glyne & Heather Newell ms (2006). Syllabification, stress and derivation by phase in Ojibwa. Ms., McGill University, Montr€al. Scheer, Tobias 2004. A Lateral Theory of Phonology. Vol.1: What is CVCV, and why should it be? Berlin: Mouton de Gruyter. Scheer, Tobias 2005. Slavic Vowel-Zero Alternations and Government Phonology: Two Approaches, One Solution. Formal Approaches to Slavic Linguistics 13: The South Carolina Meeting, edited by Steven Franks, Frank Gladney & Mila Tasseva-Kurktchieva, 300-311. Ann Arbor: Michigan Slavic Publications.
Phase theory
==> no interpretation, no phase
